| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Thoracic and Cardiovascular Surgery, Rikshospitalet University Hospital, Oslo, Norway
b Unit of Epidemiology and Biostatistics, Center of Clinical Reasearch, Ullevål University Hospital, Oslo, Norway
c Department of Cardiac Surgery, Feiring Heart Clinic, Feiring, Norway
Accepted for publication February 12, 2010.
* Address correspondence to Dr Risnes, Department of Thoracic and Cardiovascular Surgery, Rikshospitalet, N-0027 Oslo, Norway (Email: ivar.risnes{at}rikshospitalet.no).
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Methods: The study has a dual design, a case-control, and a retrospective cohort, using a source population of 18,532 consecutive patients who underwent coronary artery bypass grafting from January 1989 to December 2000. The closing date was February 1, 2008. Median follow-up was 10.3 (range 8.1 to 18.9) years. Patients with mediastinitis were compared with a random control group without mediastinitis issued from the same source population in a ratio 1:4. The crude effect of mediastinitis was estimated using rate ratio and 95% confidence limits. Adjustment for multiconfounders was done with the Cox model. A logistic model was used to pinpoint risk factors of mediastinitis. Calibration and discrimination of a prognostic model was done.
Results: One hundred seven patients (0.6%) developed mediastinitis. Diagnosis was made 12 (9 to 19) days postoperatively. Independent risk factors of mediastinitis using the logistic model were advanced age, male gender, left main stenosis, body mass index 30 kg/m2 or greater, chronic obstructive pulmonary disease, diabetes, and increased amount of blood transfusion. There was no increased risk of early mortality (odds ratio = 0.58; 95% confidence interval 0.13 to 2.61) (p = 0.48) but there was increased risk of morbidity (intraaortic balloon pump, ventricular and supraventricular arrhythmia, stroke, inotrope, and myocardial infarction). Follow-up had a median observation time of 10.3 years. Survival for patients with mediastinitis was 49.5 ± 5.0% versus 71.0 ± 2.2% for controls (p < 0.01). Analysis of specific death causes documented that cardiac deaths were significantly more frequent in mediastinitis patients than in control patients. When controlling for the confounding effect of the other variables (age, cardiopulmonary bypass time, body mass index, chronic obstructive pulmonary disease), the hazard ratio associated with mediastinitis on long-term mortality was 1.59, 95% confidence limits (1.16 and 2.70) (p = 0.003).
Conclusions: The incidence of mediastinitis in 18,532 patients undergoing coronary artery bypass grafting surgery was low. The major preventable risk factor of mediastinitis was amount of blood transfusion. Mediastinitis had an excess risk of early morbidity and was associated with a significant reduced long-term survival. Most deaths were considered to be cardiac.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
The traditional management consists of reoperation with debridement and drainage, followed by prolonged hospitalization for antibiotic therapy. Early or in-hospital mortality is reported to vary between 10 and 47% [1–7].
The etiology and pathophysiology of mediastinitis is complex and multifactorial. Previous studies have listed a number of factors associated with mediastinitis [1–10]. In the present study a case-control design was undertaken in order to evaluate preoperative, intraoperative, and postoperative risk factors of mediastinitis for patients undergoing sole CABG. As the long-term effect of mediastinitis on mortality after CABG has not been thoroughly well studied, a retrospective cohort analysis was performed to study the early and late effects of mediastinitis concerning survival.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Study Design
This study has a double design, where the first one is a case-control. Cases are patients who developed mediastinitis, and controls are patients without mediastinitis, coming from the same source population. The other is a cohort with long-term follow-up, and endpoints total mortality. In the source population mediastinitis occurred in 107 patients during the first postoperative weeks. A random sample of 444 patients was extracted from patients not suffering from mediastinitis and from the same source population. This was used as a control for the cases in a case-control design (Fig 1).
|
Statistical Methods
For the case-control study, after univariate analysis the multivariate analyses were used to pinpoint independent risk factors of mediastinitis. Association between risk factors and mediastinitis were estimated by the odds ratio (OR) and 95% confidence interval (CI). Independent risk factors were pinpointed by a backward elimination procedure, using a multivariate logistic model [13]. As we are in a prognostic strategy in the case-control analysis the predictive accuracy of the model were evaluated by calibration and discrimination.
Calibration, which measures the ability of the model to assign the appropriate risk, was evaluated by the Hosmer and Lemeshow (H-L) goodness of fit test. The H-L 
2 measures the difference between expected and observed outcomes over deciles of risk.
A statistically not significant H-L result (p value > 0.05) suggests that the model predicts accurately on average. Discrimination, which measures the ability of the model to differentiate among those who have or have not suffered mediastinitis, was evaluated by the analysis of the area under the receiver operating characteristic curve. If the area under the curve is greater than 0.7 it can be concluded that the model has an acceptable discriminatory capability.
For the cohort study the association between presence (yes/no) of mediastinitis and early mortality and morbidity was quantified using odds ratio and its 95% confidence limit (CL). In this study our strategy was explanatory, describing the crude effect of mediastinitis on long-term survival, and the adjusted effect when controlling for major confounders [15]. This is a dynamic cohort where date of surgery is time of entry in the cohort, and mortality is the major outcome. It is censored survival data where the response is mortality, and patients not dying were censored at the closing date of the study which was February 1, 2008. Long-term survival was estimated univariately by survival curves using the Kaplan-Meier method and comparison of the survival curves was done by the Breslow and Mantel-Cox test [16, 17]. A stratification analysis using the Mantel-Haenszel methods adapted to person/years was done to quantify confounders and pinpoint effect modifier using the Breslow-Day test of heterogeneity [13]. Crude effect of mediasitinitis on total mortality using the patient/years model was estimated. The adjusted effect of mediastinitis on long-term survival, controlling for multiconfounders, was done using the Cox model [17].
It is standard practice in epidemiologic research studies to compare observed survival in a cohort versus the expected survival in a reference population (the national population) of the same age and sex distribution. The method of standard mortality ratio (SMR) analysis consists of applying age and gender specific rates of death of the total Norwegian population to the cohort of heart-operated patients to yield a number of expected deaths. The SMR is the ratio of total observed deaths in the cohort divided with the total of expected deaths in the same population. The 95% confidence limit is estimated by the method of Rothman and colleagues [14]. In most instances the SMR should be regarded with caution when the expected value is less than 2. Such results could be misleading except when the disease is rare in the nonexposed population.
Power Analysis: Case-Control Design
A prestudy power analysis was done using estimates from the literature [1, 2, 7]. One of the hypotheses concerned obesity, defined as body mass index (BMI) 30 kg/m2 or greater. The frequency of obesity was 3% in patients without mediastinitis. Patients with mediastinitis had a 3.6 higher risk of being exposed to obesity. This permitted us to estimate a priori power for our study.
Considering four controls per case of mediastinitis and a type I error of 5% and a power of 80%, we would need a minimum of 103 cases of mediastinitis and 412 controls. This sample size would provide high power to pinpoint a large spectrum of independent risk factors [14, 15].
Power Analysis: Cohort Design
The power analysis was done at the planning phase of the study. We used the results from the studies of Milano and colleagues [1], where they stipulated a 2-year cumulative mortality in patients without mediastnitis to 10.1% and 23% for patients with mediastinitis. For a relative risk = 2, type I error of 5%, power of 80%, and a ratio of nonexposed to exposed of 4:1 we would need a total of 115 cases with exposed mediastinitis and 460 patients not exposed to mediastinitis.
Surgical Technique in the Primary Heart Operation
In all patients the primary operative approach was a median sternotomy, with cardiopulmonary bypass and systemic moderate hypothermia, using antegrade crystalloid cardioplegia. Standard technique included routine use of the left internal mammary artery, with supplemental vein grafts to obtain complete revascularization. Mediastinal shed blood was retransfused. The patients received standard antibiotic prophylactics with four doses of intravenous cephalotin (2 g) for at least 24 hours or until all drains or monitor lines had been removed.
Surgical Revision of Mediastinitis
All patients were treated with debridement, irrigation, and rewiring of the sternum, as described by Robicsek and colleagues [18]. Debridement was followed by continuous irrigation with antibiotic solution until the effluent drainage had been free from bacterial growth. If this technique did not provide healing, plastic surgical reconstruction with a muscle flap was employed.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
|
|
As we are in a prognostic strategy the calibration and discrimination of the model are important issues. The calibration has been assessed by plotting the observed proportion of events against the predicted probabilities by groups defined by predicted risks. Goodness of fit was tested with the H-L test (2 = 1.35, number of groups 7, p value = 0.9871). The H-L test was not significant, indicating a useful goodness of fit. These results indicate that the model predict accurately, both on average and across the ranges of patients, deciles of risk; hence, it is suitable for use in all (low to high-risk patients).
The risk of mediastinitis model developed demonstrates an acceptable discriminatory power as area under the receiver operating characteristic curve was 0.7498, 95% CL (0.711 and 0.785) (Fig 2).
|
|
Two patients died within 30 days; one from myocardial infarction on the 17th postoperative day and the other of sepsis and disseminated intravascular coagulation 25 days after the primary CABG operation. According to the National Death Index all the 14 control patients in the early mortality group died of ischemic heart disease.
Postoperative Bacterial Findings
At revision for mediastinitis, cultures were taken. Bacterial growth was positive in 85 patients (79.4%). The results of our bacterial tests performed are summarized in Table 4.
|
|
|
|
Cause of Mortality
The National Office of Statistics was used to assess specific cause of death for the patient population. Cardiac death was significantly more frequent in the mediastinitis group compared with the nonmediastinitis group. Cardiac death after 10 years for patients with mediastinitis was 34.6% compared with 21.4% in nonmediastinitis patients (p < 0.006). In the mediastinitis group cardiac-related deaths were found in 37 (34.6%) patients (ischemic heart disease 35, valvular disease 1 and arrhythmia 1) and noncardiac-related deaths in 19 (17.8%) patients (malignant disease, 8; diabetes mellitus, 5; chronic obstructive pulmonary disease, 3; and others, 3). In the nonmediastinitis group, 95 (21.4%) patients died of cardiac diseases (ischemic heart disease, 91; valvular disease, 2; heart insufficiency, 2) and noncardiac-related deaths in 70 (15.8%) patients (malignant disease, 25; cerebral disease, 13; diabetes mellitus, 7; COPD 4; renal insufficiency, 4; pneumonia, 3; gastrointestinal ischemia, 3; abdominal aortic aneurysm, 3; and others, 8) (p = 0.72). Cardiac death was significantly higher for the mediastinitis group compared with the nonmediastinitis (p = 0.006).
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
30 kg/m2, COPD, and age) have often been associated with occurrence of mediastinitis [1–10]. In addition we found male gender and LMS to be significantly associated with mediastinitis. Diabetes mellitus turned out to be one of the most important predictors of mediastinitis. This is in agreement with previous reports [5, 7, 19], although in contrast to others [1–3, 6, 8]. Obesity has been identified as the most important independent predictor of mediastinitis [1, 3, 4, 7]. Our study confirmed the significance of obesity.
We identified COPD disease to be an independent risk factor of mediastinitis in line with other studies [4, 20]. Smoking history was not statistically significant either in univariate or in multivariate analysis, although respiratory distress and pneumonia may occur more frequently in these patients [2, 3, 7]. Age was a significant risk factor for mediastinitis in our study in contrast to some other studies [21, 22]. Cardiopulmonary bypass time was a risk factor in the univariate analysis, but not in the multivariate analysis. The bypass time has been found important in some other reports [1, 3, 20].
There is a link between the length of preoperative hospital stay, and colonization with hospital bacterial flora and inoculation of microorganisms into cutaneous and subcutaneous tissue [23, 24]. We could not detect this association, probably because of very short preoperative hospitalization periods.
The strongest independent risk factor of mediastinitis in our study was male gender. Men had significantly higher rates of diabetes then women. This is in accordance with some other publications [5, 25]. The mechanisms by which men are more predisposed to mediastinitis than women are not known and have to be speculative. Anatomically, women differ from men by having a better collateral flow to the sternum after internal mammary artery harvest [3, 19, 26]. In addition, men have more hair follicles in the area for sternotomy, which may dispose for bacterial growth and infection.
Left main stenosis turned out to be an independent risk factor. This finding is new, and difficult to explain. The amount of blood transfusion (over 10 units) was the only independent postoperative risk factor. This may be explained by a decrease in immune function after homologous blood transfusion [3, 23, 27].
Our results underline the need for preventive measures. Measures include weight reduction (in case of elective surgery), weight adapted perioperative antibiotic treatment, and a strict antidiabetic regimen.
Mediastinitis and Early Morbidity and Long-Term Survival
In the present study mediastinitis after CABG surgery was associated with a marked increase in morbidity and long-term mortality rates. Strategies for reduction of mediastinitis included the use of an antimicrobial prophylaxis standard aseptic surgical technique, attention to hemostasis, and precise sternal closure, as also underlined in previous reports [1–10].
Hospital mortality of mediastinitis is frequently related to uncontrolled infection, sepsis, and multiorgan failure. Early diagnosis of mediastinitis may improve short-term results, in particular by preventing development of these devastating complications. We report that mediastinitis did not increase early mortality. This could be due to power deficiency as far this endpoint in our study.
The observed effect of mediastinitis on true long-term survival is novel information. Post-CABG mediastinitis is associated with a substantial lower survival compared with patients without mediastinitis.
This study has a median follow-up of 10.3 years, and a range of less than one year to a maximum of 18.9 years. This was the longest follow-up time compared with other studies. Milano and colleagues [1] had a median follow-up of two years, and Loop and colleagues [3]) had a median follow-up of 2.1 years. Finally, Braxton and colleagues [4] had a mean follow-up of 4.1 years, while Ståhle and colleagues [6] had a mean follow-up of 5 years. Without any doubt, to our knowledge, our longitudinal cohort study has the longest median follow-up time of all studies published until now.
On the other hand there is a concordance in the estimated hazard of mediastinitis on mortality between our study and the others. Our crude rate ratio of the effect of mediastinitis on mortality was 2.01, 95% CL (1.49 and 2.74). Milano and colleagues [1] had 2.2 to 2 years, while Loop and colleagues [3] had a rate ratio of 3.4 at 3 years and Ståhle and colleagues [6] had a rate ratio of 2.4 at 5 years follow-up.
Also, there was a concordance in the adjusted hazard ratio for multiconfounders estimate. We had a hazard ratio of 1.59, 95% CL (1.16 and 2.55). This was close to the adjusted hazard rate of Ståhle and colleagues [6] of 1.90 (1.20 and 2.80), and to the estimate of Braxton and colleagues [4] 1.79 95% CL (1.30 and 2.19). The replication of the increased risk of mediastinitis on mortality in different geographic populations of CABG patients, and in different periods of time, is an indicator of the validity of this association.
Another original result confirming the consistency of this association is the finding that SMR for males with mediastinitis had an increased risk of mortality compared with males without mediastinitis when controlling for the total Norwegian population. The SMR for males without mediastinitis of 2.5 ,95% CL (2.1 and 3.05) was the same as the SMR estimated by Risum and colleagues [28] (2.5, 95% CL [2.2 and 2.9] for males, considering a cohort of 1,025 patients operated between August 1982 to December 1986 (another time period) with a median follow-up of 7.4 years in Norway. These analyses were impossible to perform for females in our study.
There are some biologic mechanisms that can explain this excess risk of mediastinitis. Information is scarce with regard to the long-term graft patency in patients who develop mediastinitis. These patients have a long-standing chronic inflammatory process, which might influence negatively the graft patency. Inflammation can reduce plaque stability and increase thrombogenesis in both coronaries and grafts [29, 30].
Hypercholesterolemia may also play a role by activating the inflammatory reaction associated with accelerated progression of atherosclerosis [22]. These factors might contribute to explain the reduced long-term survival rate in CABG patients having mediastinitis.
Deep sternal wound infections may cause constrictive pericarditis. Even patients with mild constriction have a comprised left ventricular filling and decreased stroke volume and cardiac output. Pericardial, as well as epicardial, fibrosis after mediastinitis is thus another possible contributor to the observed excess long-term mortality [22, 29, 30].
Mediastinitis is a devastating complication after cardiac surgery and is associated with a significant reduced long-term survival. At 10.3 years follow-up these patients had a 59% higher mortality risk compared with the patients without mediastinitis, and mostly caused by cardiac deaths.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
lu D, Omerolu SN, Kaya E, et al. Does mediastinitis affect the graft patency J Card Surg 2005;20:208-211.[Medline]Related Article
This article has been cited by other articles:
|
|
 |
|
  H. Valenzuela, Y. Carrascal, L. Maroto, and N. Arce Fulminant mediastinitis due to extended-spectrum beta-lactamase-producing Klebsiella pneumoniae: atypical presentation and spreading following cardiac surgery Interact CardioVasc Thorac Surg, May 1, 2013; 16(5): 703 - 704. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Melly, B. Gahl, R. Meinke, F. Rueter, P. Matt, O. Reuthebuch, F. S. Eckstein, and M. T. R. Grapow A new cable-tie-based sternal closure device: infectious considerations Interact CardioVasc Thorac Surg, April 26, 2013; (2013) ivt183v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. G. Koch, E. R. Nowicki, J. Rajeswaran, S. M. Gordon, J. F. Sabik III, and E. H. Blackstone When the timing is right: Antibiotic timing and infection after cardiac surgery J. Thorac. Cardiovasc. Surg., October 1, 2012; 144(4): 931 - 937.e4. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Kamiya, S. S. A. Al-maisary, P. Akhyari, A. Ruhparwar, K. Kallenbach, A. Lichtenberg, and M. Karck The number of wires for sternal closure has a significant influence on sternal complications in high-risk patients Interact CardioVasc Thorac Surg, October 1, 2012; 15(4): 665 - 670. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Osada, H. Nakajima, M. Morishima, and T. Su Candidal mediastinitis successfully treated using vacuum-assisted closure following open-heart surgery Interact CardioVasc Thorac Surg, June 1, 2012; 14(6): 872 - 874. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Hajj-Chahine, J. Tomasi, H. Houmaida, and E. Bonnin eComment: Do gentamicin-collagen sponges decrease the risk of mediastinitis? Interact CardioVasc Thorac Surg, April 1, 2012; 14(4): 394 - 394. [Full Text] [PDF]
|
|
|
|
|
|
R. J. Vos, A. Yilmaz, U. Sonker, J. C. Kelder, and G. T. L. Kloppenburg Vacuum-assisted closure of post-sternotomy mediastinitis as compared to open packing Interact CardioVasc Thorac Surg, January 1, 2012; 14(1): 17 - 21. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. J. Tuyl, J. H. Mackney, and C. L. Johnston Management of Sternal Precautions Following Median Sternotomy by Physical Therapists in Australia: A Web-Based Survey Physical Therapy, January 1, 2012; 92(1): 83 - 97. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Writing Committee Members, L. D. Hillis, P. K. Smith, J. L. Anderson, J. A. Bittl, C. R. Bridges, J. G. Byrne, J. E. Cigarroa, V. J. DiSesa, L. F. Hiratzka, et al. 2011 ACCF/AHA Guideline for Coronary Artery Bypass Graft Surgery: A Report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines Circulation, December 6, 2011; 124(23): e652 - e735. [Full Text] [PDF]
|
|
|
|
 |
|
 T. C. Berg, K. E. Kjorstad, P. E. Akselsen, B. E. Seim, H. L. Lower, M. N. Stenvik, N. K. Sorknes, and H.-M. Eriksen National surveillance of surgical site infections after coronary artery bypass grafting in Norway: incidence and risk factors Eur J Cardiothorac Surg, December 1, 2011; 40(6): 1291 - 1297. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Hajj-Chahine, C. Jayle, J. Tomasi, H. Houmaida, and P. Corbi Same Admission Cardiac Catheterization-Cardiac Surgery: Increased Incidence of Acute Kidney Injury and Mediastinitis Ann. Thorac. Surg., August 1, 2011; 92(2): 776 - 776. [Full Text] [PDF]
|
|
|
|
|
|
W. L. Holman Invited Commentary Ann. Thorac. Surg., May 1, 2010; 89(5): 1510 - 1510. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
